Pathways for optimising colorectal cancer screening programs: a modelling assessment

defence of my doctoral thesis

Thursday 22 April 2021

10:30am

In light of the COVID-19

pandemic, the defence

ceremony will be held online.

Access to the live stream of

the ceremony will be

communicated later.

dayna.cenin@gmail.com

Paranymphs

Nadine Zielonke

Clare Aitken

defence of my doctoral thesis

Thursday 22 April 2021

10:30am

In light of the COVID-19

pandemic, the defence

ceremony will be held online.

Access to the live stream of

the ceremony will be

communicated later.

dayna.cenin@gmail.com

Paranymphs

Nadine Zielonke

Clare Aitken

defence of my doctoral thesis

Thursday 22 April 2021

10:30am

In light of the COVID-19

pandemic, the defence

ceremony will be held online.

Access to the live stream of

the ceremony will be

communicated later.

dayna.cenin@gmail.com

Paranymphs

Nadine Zielonke

Clare Aitken

A MODELLING ASSESSMENT

Pathways for Optimising Colorectal Cancer Screening Programs: a modelling assessment Dayna Rene Cenin

Doctoral thesis, Erasmus University Rotterdam, the Netherlands

This thesis was financially supported by the Department of Public Health, Erasmus MC University Medical Center Rotterdam and the Erasmus University Rotterdam.

ISBN: 978-94-6332-750-3

Printing: GVO drukkers & vormgevers Cover design: Kalinga White

©Dayna Rene Cenin, 2021

All right reserved. No part of this thesis may be reproduced in any form, by print, photocopy, digital file, internet or any other means without permission from the author or the

copyright-owning journals for previously published chapters.

A MODELLING ASSESSMENT

Strategieën om darmkanker screening te optimaliseren een model evaluatie

Proefschift

to obtain the degree of Doctor from the Erasmus University Rotterdam

by command of the rector magnificus

Prof. Dr. F.A. van der Duijn Schouten

and in accordance with the decision of the Doctorate Board. The public defence shall be held on

Thursday 22 April 2021 at 10.30hrs by

Dayna Rene Cenin

Pathways for Optimising Colorectal Cancer Screening Programs: a modelling assessment Dayna Rene Cenin

Doctoral thesis, Erasmus University Rotterdam, the Netherlands

This thesis was financially supported by the Department of Public Health, Erasmus MC University Medical Center Rotterdam and the Erasmus University Rotterdam.

ISBN: 978-94-6332-750-3

Printing: GVO drukkers & vormgevers Cover design: Kalinga White

©Dayna Rene Cenin, 2021

All right reserved. No part of this thesis may be reproduced in any form, by print, photocopy, digital file, internet or any other means without permission from the author or the

copyright-owning journals for previously published chapters.

A MODELLING ASSESSMENT

Strategieën om darmkanker screening te optimaliseren een model evaluatie

Proefschift

to obtain the degree of Doctor from the Erasmus University Rotterdam

by command of the rector magnificus

Prof. Dr. F.A. van der Duijn Schouten

and in accordance with the decision of the Doctorate Board. The public defence shall be held on

Thursday 22 April 2021 at 10.30hrs by

Dayna Rene Cenin

Promotor:

Prof. Dr. H.J. de Koning Overige leden:

Prof. Dr. V.M.C.W. Spaander Prof. Dr. A.G. Uitterlinden Prof. N. Pashayan Copromotor:

Dr. I. Lansdorp-Vogelaar

For Nor Nor You always believed I could do anything.

Promotor:

Prof. Dr. H.J. de Koning Overige leden:

Prof. Dr. V.M.C.W. Spaander Prof. Dr. A.G. Uitterlinden Prof. N. Pashayan Copromotor:

Dr. I. Lansdorp-Vogelaar

For Nor Nor You always believed I could do anything.

General Introduction 9

Part 1: Optimisation of uniform screening programs 39

Chapter 1: Costs and outcomes of Lynch syndrome screening in the Australian colorectal cancer

population 41

Chapter 2: Optimising colorectal cancer screening in Shanghai, China: a modelling study 63

Chapter 3: Optimising the expansion of the National Bowel Cancer Screening Program 115

Part 2: Optimisation through personalisation 135

Chapter 4: Colorectal cancer screening with faecal immunochemical testing, sigmoidoscopy or

colonoscopy: a microsimulation modelling study 137

Chapter 5: Calculation of stop ages for colorectal cancer screening based on comorbidities and

screening history 197

Chapter 6: Cost-effectiveness of personalised screening for colorectal cancer based on polygenic

risk and family history 245

General Discussion 295

Model Appendix 335

Summary 361

Samenvatting 371

About the author 381

General Introduction 9

Part 1: Optimisation of uniform screening programs 39

Chapter 1: Costs and outcomes of Lynch syndrome screening in the Australian colorectal cancer

population 41

Chapter 2: Optimising colorectal cancer screening in Shanghai, China: a modelling study 63

Chapter 3: Optimising the expansion of the National Bowel Cancer Screening Program 115

Part 2: Optimisation through personalisation 135

Chapter 4: Colorectal cancer screening with faecal immunochemical testing, sigmoidoscopy or

colonoscopy: a microsimulation modelling study 137

Chapter 5: Calculation of stop ages for colorectal cancer screening based on comorbidities and

screening history 197

Chapter 6: Cost-effectiveness of personalised screening for colorectal cancer based on polygenic

risk and family history 245

295 335 361 371 381 General Discussion Model Appendix Summary Samenvatting About the author

C

C

:

A

G

P

H

P

E

INCIDENCE

Colorectal cancer is an important global public health issue. With over 1.8 million new diagnoses in 2018, colorectal cancer was the third leading cause of cancer incidence,

representing approximately 10% of the global cancer burden (Figure 1a).1-3 _{Colorectal cancer}

incidence increases steeply with age, especially in those aged above 50 year and the median

age at diagnosis is 66-70 years in developed countries.4, 5 _{At younger ages, colorectal cancer is}

rare and is generally associated with inherited genetic mutations.6 _{Nonetheless, there is a}

growing body of evidence that colorectal cancer incidence in those under 50 is increasing.7-10

Incidence is consistently higher in males than in females,1-3 _{however globally there is}

significant variation (Figure 2).1-3 _{In 2018, the estimated age-standardised incidence by region}

ranged from as low as 1.7 cases per 100,000 for males in Africa to 70.6 per 100,000 in Europe, and from 0.5 to 39.3 per 100,000 in the same regions for females.

Traditionally thought of as a disease of the “western world”, incidence of colorectal cancer is rising in populations historically considered to be at low risk. This change is largely a result of temporal trends such as population ageing and improved standards of living, which has resulted in the adoption of the Western lifestyle including changes in dietary habits and a rise in modifiable risk factors such as smoking, alcohol consumption, obesity, and lack of physical

activity in these populations.11-14 _{Conversely, incidence is stabilising or declining in many high}

income countries, partly due to the implementation of screening programs.14, 15

Although individuals in westernised countries generally experience greater risk of colorectal

cancer (Figure 2), globally the overall burden of colorectal cancer is unevenly distributed.1-3

Increased incidence, coupled with large population size means that countries like China are noteworthy contributors to the global burden. In 2018, China accounted for approximately 28.2% of colorectal cancer cases. In comparison, Australia and the Netherlands contribute just 1.0% and 0.8% respectively.

MORTALITY

In 2018, more than 880,000 individuals died from colorectal cancer (Figure 1b), making it the

second leading cause of cancer-related deaths.1-3 _{This is despite increasing awareness of the}

disease and its impact among researchers, policymakers and the general public. Although mortality is considerably lower than incidence, like incidence, the burden is unevenly

distributed with wide ranges in estimated age-standardised mortality.1-3 _{Despite higher}

In

tr

oduction

C

C

:

A

G

P

H

P

E

INCIDENCE

Colorectal cancer is an important global public health issue. With over 1.8 million new diagnoses in 2018, colorectal cancer was the third leading cause of cancer incidence,

representing approximately 10% of the global cancer burden (Figure 1a).1-3 _{Colorectal cancer}

incidence increases steeply with age, especially in those aged above 50 year and the median

age at diagnosis is 66-70 years in developed countries.4, 5 _{At younger ages, colorectal cancer is}

rare and is generally associated with inherited genetic mutations.6 _{Nonetheless, there is a}

growing body of evidence that colorectal cancer incidence in those under 50 is increasing.7-10

Incidence is consistently higher in males than in females,1-3 _{however globally there is}

significant variation (Figure 2).1-3 _{In 2018, the estimated age-standardised incidence by region}

ranged from as low as 1.7 cases per 100,000 for males in Africa to 70.6 per 100,000 in Europe, and from 0.5 to 39.3 per 100,000 in the same regions for females.

Traditionally thought of as a disease of the “western world”, incidence of colorectal cancer is rising in populations historically considered to be at low risk. This change is largely a result of temporal trends such as population ageing and improved standards of living, which has resulted in the adoption of the Western lifestyle including changes in dietary habits and a rise in modifiable risk factors such as smoking, alcohol consumption, obesity, and lack of physical

activity in these populations.11-14 _{Conversely, incidence is stabilising or declining in many high}

income countries, partly due to the implementation of screening programs.14, 15

Although individuals in westernised countries generally experience greater risk of colorectal

cancer (Figure 2), globally the overall burden of colorectal cancer is unevenly distributed.1-3

Increased incidence, coupled with large population size means that countries like China are noteworthy contributors to the global burden. In 2018, China accounted for approximately 28.2% of colorectal cancer cases. In comparison, Australia and the Netherlands contribute just 1.0% and 0.8% respectively.

MORTALITY

In 2018, more than 880,000 individuals died from colorectal cancer (Figure 1b), making it the

second leading cause of cancer-related deaths.1-3 _{This is despite increasing awareness of the}

disease and its impact among researchers, policymakers and the general public. Although mortality is considerably lower than incidence, like incidence, the burden is unevenly

distributed with wide ranges in estimated age-standardised mortality.1-3 _{Despite higher}

a

b

Figure 1: Estimated worldwide cancer a) incidence and b) mortality for males and females of

all ages in 2018.1-3

a

b

Figure 2: Estimated Age-standardised incidence rates (World) for a) females and b) males of

In

tr

oduction

a

b

Figure 1: Estimated worldwide cancer a) incidence and b) mortality for males and females of

all ages in 2018.1-3

a

b

Figure 2: Estimated Age-standardised incidence rates (World) for a) females and b) males of

deaths in 2018. In the same period, China, due to its large population, contributed a significantly greater proportion accounting for approximately 28.1% of colorectal cancer deaths.

Temporal trends show that mortality is increasing in countries like China, while in more

developed countries mortality rates are declining.14 _{This decline in mortality in developed}

countries is multifaceted and can be attributed to improved survival due to earlier diagnosis

of colorectal cancer as a result of screening16, 17 _{and advances in our understanding of the}

disease. This has resulted in improved surgical and adjuvant therapy and the adoption of best

practice in cancer treatment and management of colorectal cancer.18-20 _{Increases in mortality,}

in countries like China may reflect issues with health infrastructure, disparities in access to

cancer care and limited or no access to early detection and appropriate treatment.21, 22

Given demographic projections and ongoing societal and economic developments in many low-and middle-income countries the global burden of colorectal cancer is expected to increase dramatically. By 2040, colorectal cancer incidence is projected to increase by 70% to more than 3.1 million new cases, while mortality will increase by more than 80% to an

estimated 1.6 million colorectal cancer deaths.2, 23, 24

N

Colorectal cancer develops from benign precursor lesions or polyps in the colorectum, the final section of the gastrointestinal tract that performs the vital task of absorbing water and nutrients while converting digested food into faeces. Until recently, it was believed that colorectal cancer only developed from a lesion or polyp known as an adenoma, in what is known as the adenoma-carcinoma sequence. However, it is now recognised that this so-called

‘conventional pathway’ is only one way in which colorectal cancer develops.25-27 _Serrated

polyps are now considered an important premalignant lesion with colorectal cancer developing through the serrated polyp pathway. Although there remains debate about the magnitude of the impact of this pathway, estimates suggest it accounts for between 15-30%

of colorectal cancers.26, 27

Conventional adenomas and serrated lesions can vary in size and conventional adenoma can also vary in shape (pedunculated (stalked), elevated, flat, oblong or depressed). Serrated lesions can be divided in three subgroups: hyperplastic polyps, sessile serrated polyps and traditional serrated polyps. Hyperplastic polyps are thought to be non-malignant and thus will not develop into colorectal cancer. Sessile serrated polyps and traditional serrated polyps are histologically distinguishable from conventional adenomas by their saw-tooth configuration. Data on the natural history of this pathway is limited and the lesions can be difficult to detect

during colonoscopy due to their discrete appearance and frequent covering of mucous.26, 28

Figure 3: Traditional Adenoma-Carcinoma sequence. (image courtesy of BMJ)29

Adenomas are common – approximately 20-50% of the population will develop one or more

adenomas during their lifetime.26 _{Despite the high risk of developing an adenoma, lifetime risk}

of developing colorectal cancer is much lower, ranging from 5-8% in westernised countries.5,

26, 30 _{Identifying those lesions which are truly at risk of developing into cancer remains a serious}

and clinically relevant challenge, as it is necessary to avoid both overtreatment and

undertreatment.28 _{To date, serrated lesions have been over-represented in interval cancers.}31,

32

High risk characteristics of polyps include size (diameter >1cm), number, proximal location,

and the presence of dysplasia.26, 33 _{For adenomas, histology (tubular or villous) is also}

important while for serrated lesions, molecular features, such as methylation, BRAF gene

mutations and micro-satellite instability also increase risk.32

The transition from a small polyp into colorectal cancer is characterised by a multistep process which involves a series of histological, morphological and genetic changes over time. Although the life history of the sequence is likely to be highly variable, the progression from an asymptomatic polyp or lesion to a symptom detectable colorectal cancer takes on average

10-15 years.34 _{Microsimulation models suggest this dwell time to be between 10-25 years.}35

Encouragingly, colorectal cancer is largely preventable if it, or its precursor lesion, is detected and treated in the early stages. The long latent period provides a substantial window of opportunity to achieve this (see Secondary Prevention).

As lesions progress, symptoms may become present but they are often non-specific in nature:

abdominal pain, change in bowel habits, rectal blood loss, or weight loss.36 _{By the time the}

signs of colorectal cancer become apparent, the disease has usually progressed to an advanced stage.

In

tr

oduction

deaths in 2018. In the same period, China, due to its large population, contributed a significantly greater proportion accounting for approximately 28.1% of colorectal cancer deaths.

Temporal trends show that mortality is increasing in countries like China, while in more

developed countries mortality rates are declining.14 _{This decline in mortality in developed}

countries is multifaceted and can be attributed to improved survival due to earlier diagnosis

of colorectal cancer as a result of screening16, 17 _{and advances in our understanding of the}

disease. This has resulted in improved surgical and adjuvant therapy and the adoption of best

practice in cancer treatment and management of colorectal cancer.18-20 _{Increases in mortality,}

in countries like China may reflect issues with health infrastructure, disparities in access to

cancer care and limited or no access to early detection and appropriate treatment.21, 22

Given demographic projections and ongoing societal and economic developments in many low-and middle-income countries the global burden of colorectal cancer is expected to increase dramatically. By 2040, colorectal cancer incidence is projected to increase by 70% to more than 3.1 million new cases, while mortality will increase by more than 80% to an

estimated 1.6 million colorectal cancer deaths.2, 23, 24

N

Colorectal cancer develops from benign precursor lesions or polyps in the colorectum, the final section of the gastrointestinal tract that performs the vital task of absorbing water and nutrients while converting digested food into faeces. Until recently, it was believed that colorectal cancer only developed from a lesion or polyp known as an adenoma, in what is known as the adenoma-carcinoma sequence. However, it is now recognised that this so-called

‘conventional pathway’ is only one way in which colorectal cancer develops.25-27 _Serrated

polyps are now considered an important premalignant lesion with colorectal cancer developing through the serrated polyp pathway. Although there remains debate about the magnitude of the impact of this pathway, estimates suggest it accounts for between 15-30%

of colorectal cancers.26, 27

Conventional adenomas and serrated lesions can vary in size and conventional adenoma can also vary in shape (pedunculated (stalked), elevated, flat, oblong or depressed). Serrated lesions can be divided in three subgroups: hyperplastic polyps, sessile serrated polyps and traditional serrated polyps. Hyperplastic polyps are thought to be non-malignant and thus will not develop into colorectal cancer. Sessile serrated polyps and traditional serrated polyps are histologically distinguishable from conventional adenomas by their saw-tooth configuration. Data on the natural history of this pathway is limited and the lesions can be difficult to detect

during colonoscopy due to their discrete appearance and frequent covering of mucous.26, 28

Figure 3: Traditional Adenoma-Carcinoma sequence. (image courtesy of BMJ)29

Adenomas are common – approximately 20-50% of the population will develop one or more

adenomas during their lifetime.26 _{Despite the high risk of developing an adenoma, lifetime risk}

of developing colorectal cancer is much lower, ranging from 5-8% in westernised countries.5,

26, 30 _{Identifying those lesions which are truly at risk of developing into cancer remains a serious}

and clinically relevant challenge, as it is necessary to avoid both overtreatment and

undertreatment.28 _{To date, serrated lesions have been over-represented in interval cancers.}31,

32

High risk characteristics of polyps include size (diameter >1cm), number, proximal location,

and the presence of dysplasia.26, 33 _{For adenomas, histology (tubular or villous) is also}

important while for serrated lesions, molecular features, such as methylation, BRAF gene

mutations and micro-satellite instability also increase risk.32

The transition from a small polyp into colorectal cancer is characterised by a multistep process which involves a series of histological, morphological and genetic changes over time. Although the life history of the sequence is likely to be highly variable, the progression from an asymptomatic polyp or lesion to a symptom detectable colorectal cancer takes on average

10-15 years.34 _{Microsimulation models suggest this dwell time to be between 10-25 years.}35

Encouragingly, colorectal cancer is largely preventable if it, or its precursor lesion, is detected and treated in the early stages. The long latent period provides a substantial window of opportunity to achieve this (see Secondary Prevention).

As lesions progress, symptoms may become present but they are often non-specific in nature:

abdominal pain, change in bowel habits, rectal blood loss, or weight loss.36 _{By the time the}

signs of colorectal cancer become apparent, the disease has usually progressed to an advanced stage.

S

Survival from colorectal cancer is highly dependent on the stage of tumour development at diagnosis, as well as the treatments that follows. Although survival has steadily improved

during the past decades in many countries,37 _{overall five year survival estimates vary}

noticeably around the world; reaching between 60-70% in high-income countries, such as

Australia,5 _{Canada, the USA,}4, 38 _{and several European countries,}37, 39, 40 _{while remaining below}

50% in low-income settings, including China.41 _{However, it is clear that improved survival is}

expected with earlier detection and diagnosis of colorectal cancer (Table 1).5, 38, 39 _The

detection of colorectal cancer before it develop into its later stages has a profound effect on mortality; significantly fewer patients with late stage colorectal cancer survive five years post diagnosis, compared to those diagnosed at earlier stages.

Table 1: Estimate of five-year survival rate post diagnosis at each stage in the USA, Australia and the Netherlands.

Stage USA38 _Australia5 _Netherlands39

I 88 99 94

II 80 89 85

III 66 71 72

IV 13 13 12

Unknown 39 57 29

Early detection and diagnosis also improve treatment outcomes and lessens the need for expensive and invasive therapies. Left in situ in patients who refuse polypectomy, adenomas

are more likely to go on and develop colorectal cancer42 _{while the removal of adenomatous}

polyps, coupled with regular and ongoing surveillance, has been shown to reduce the risk of

colorectal cancer.43

Further improvements in survival will occur as a result of earlier detection through organised screening of asymptomatic individuals and this will have a significant impact on five-year survival rates. In addition, advances in colorectal cancer treatment will continue to make incremental improvements in survival but this will likely be at a considerable financial cost.

A

C

C

There is no single risk factor or cause that accounts for the majority of cases of colorectal cancer. Rather, there are a range of factors that are known to effect risk. Encouragingly, many of the factors that increase an individual’s risk are, in principle, modifiable. These risk factors

are common in nature and therefore account for a larger proportion of the disease burden at the population-level, despite lower relative risks.

M

Modifiable risk factors are things that an individual can change in order to reduce their risk of a given disease. In relation to colorectal cancer these modifiable risk factors include diet,

alcohol consumption, smoking, obesity, and physical (in)activity.44-47

There is convincing evidence that consumption of red and processed meats increases risk of

both adenomas and colorectal cancer.48-51 _{In addition, consumption of alcoholic drinks}52 _and

cigarette smoking53 _{are also convincingly associated with an increased risk for colorectal}

cancer. A clear dose response mechanism has been established for each of these risk factors, highlighting that as consumption of these substances increases, so too does the risk of colorectal cancer. In addition, there is a well-established link between obesity (body and

abdominal fatness) and increased colorectal cancer risk,54 _{such that those with the largest}

body mass index or waist circumference are at greatest risk.

In contrast, participation in all types of physical activity has been convincingly shown to reduce

the risk of colon cancer, although no conclusion has been drawn for rectal cancer.55 _{Like those}

factors that increase colorectal cancer risk, there is a clear dose response relationship with those participating in the highest level of activity being at significantly lower risk compared

with those who are least active. In addition, dietary fibre56 _{and intake of calcium (either}

through supplementation57 _{or consumption of milk and other dairy products}58_{) are considered}

to be protective and are associated with probable reduction in colorectal cancer risk.

N

-

There are a range of non-modifiable factors that increase an individual’s colorectal cancer risk.

The most well-known non-modifiable risk factors are male sex and advancing age.1, 59 _In

addition, several diseases can lead to an increased colorectal cancer risk. These include

inflammatory bowel diseases (Crohn’s disease60 _{and ulcerative colitis}61_{) type II diabetes}62, 63

and cystic fibrosis.64, 65 _{There is emerging evidence to suggest that infection with Helicobacter}

pylori66 _{and Fusobacterium nucleatum}67 _{may also be associated with an increased risk of}

colorectal cancer.

Aside from the above factors, possibly the most important non-modifiable risk factor for colorectal cancer is family history. Individuals with a positive family history, but without an identified cancer syndrome, are at increased risk for developing the disease. The risk increases with the number of relatives diagnosed with colorectal cancer, the closeness of genetic relationship of the diagnosed relative(s), and the age of diagnosis of the relative(s). For example, for those who have a first-degree relative with colorectal cancer diagnosed at an

In

tr

oduction

S

Survival from colorectal cancer is highly dependent on the stage of tumour development at diagnosis, as well as the treatments that follows. Although survival has steadily improved

during the past decades in many countries,37 _{overall five year survival estimates vary}

noticeably around the world; reaching between 60-70% in high-income countries, such as

Australia,5 _{Canada, the USA,}4, 38 _{and several European countries,}37, 39, 40 _{while remaining below}

50% in low-income settings, including China.41 _{However, it is clear that improved survival is}

expected with earlier detection and diagnosis of colorectal cancer (Table 1).5, 38, 39 _The

detection of colorectal cancer before it develop into its later stages has a profound effect on mortality; significantly fewer patients with late stage colorectal cancer survive five years post diagnosis, compared to those diagnosed at earlier stages.

Table 1: Estimate of five-year survival rate post diagnosis at each stage in the USA, Australia and the Netherlands.

Stage USA38 _Australia5 _Netherlands39

I 88 99 94

II 80 89 85

III 66 71 72

IV 13 13 12

Unknown 39 57 29

Early detection and diagnosis also improve treatment outcomes and lessens the need for expensive and invasive therapies. Left in situ in patients who refuse polypectomy, adenomas

are more likely to go on and develop colorectal cancer42 _{while the removal of adenomatous}

polyps, coupled with regular and ongoing surveillance, has been shown to reduce the risk of

colorectal cancer.43

Further improvements in survival will occur as a result of earlier detection through organised screening of asymptomatic individuals and this will have a significant impact on five-year survival rates. In addition, advances in colorectal cancer treatment will continue to make incremental improvements in survival but this will likely be at a considerable financial cost.

A

C

C

There is no single risk factor or cause that accounts for the majority of cases of colorectal cancer. Rather, there are a range of factors that are known to effect risk. Encouragingly, many of the factors that increase an individual’s risk are, in principle, modifiable. These risk factors

are common in nature and therefore account for a larger proportion of the disease burden at the population-level, despite lower relative risks.

M

Modifiable risk factors are things that an individual can change in order to reduce their risk of a given disease. In relation to colorectal cancer these modifiable risk factors include diet,

alcohol consumption, smoking, obesity, and physical (in)activity.44-47

There is convincing evidence that consumption of red and processed meats increases risk of

both adenomas and colorectal cancer.48-51 _{In addition, consumption of alcoholic drinks}52 _and

cigarette smoking53 _{are also convincingly associated with an increased risk for colorectal}

cancer. A clear dose response mechanism has been established for each of these risk factors, highlighting that as consumption of these substances increases, so too does the risk of colorectal cancer. In addition, there is a well-established link between obesity (body and

abdominal fatness) and increased colorectal cancer risk,54 _{such that those with the largest}

body mass index or waist circumference are at greatest risk.

In contrast, participation in all types of physical activity has been convincingly shown to reduce

the risk of colon cancer, although no conclusion has been drawn for rectal cancer.55 _{Like those}

factors that increase colorectal cancer risk, there is a clear dose response relationship with those participating in the highest level of activity being at significantly lower risk compared

with those who are least active. In addition, dietary fibre56 _{and intake of calcium (either}

through supplementation57 _{or consumption of milk and other dairy products}58_{) are considered}

to be protective and are associated with probable reduction in colorectal cancer risk.

N

-

There are a range of non-modifiable factors that increase an individual’s colorectal cancer risk.

The most well-known non-modifiable risk factors are male sex and advancing age.1, 59 _In

addition, several diseases can lead to an increased colorectal cancer risk. These include

inflammatory bowel diseases (Crohn’s disease60 _{and ulcerative colitis}61_{) type II diabetes}62, 63

and cystic fibrosis.64, 65 _{There is emerging evidence to suggest that infection with Helicobacter}

pylori66 _{and Fusobacterium nucleatum}67 _{may also be associated with an increased risk of}

colorectal cancer.

Aside from the above factors, possibly the most important non-modifiable risk factor for colorectal cancer is family history. Individuals with a positive family history, but without an identified cancer syndrome, are at increased risk for developing the disease. The risk increases with the number of relatives diagnosed with colorectal cancer, the closeness of genetic relationship of the diagnosed relative(s), and the age of diagnosis of the relative(s). For example, for those who have a first-degree relative with colorectal cancer diagnosed at an

early age (below age 55) or two close relatives with colorectal cancer, irrespective of the age at diagnosis, colorectal cancer risk has been found to be three- to six-fold greater than the

average population.68 _{This risk increases to seven- to ten-fold when there are at least three}

first- or second-degree relatives with colorectal cancer, with at least one diagnosed under 55 years or at least three first-degree relatives with colorectal cancer diagnosed at 55 years or older.

In addition, there are several familial genes that substantially increase an individual’s risk of early-onset colorectal cancer when they are inherited in a mutated form. These genes are associated with Lynch syndrome (previously known as hereditary non-polyposis colorectal cancer), familial adenomatous polyposis (also known as FAP) and MUTYH-associated polyposis. However, together these inherited genetic mutations account for less than 6% of

all colorectal cancer cases69, 70 _{and do not fully explain why family history is a risk for colorectal}

cancer.71, 72 _{This suggests that other factors associated with heritability may play a role,}

including dietary, lifestyle and environmental factors,73 _{mutations in yet-to-be-discovered}

colorectal cancer susceptibility genes74 _{and common, low risk variants (or single-nucleotide}

polymorphisms [SNPs]).75-77 _{Although in isolation, SNPs are only weakly associated with}

colorectal cancer risk, individuals with multiple SNPs can have a substantially increased risk. In addition, they explain substantial variation in heritability risk due to their relatively high

prevalence in the population.77-81 _{Research to date suggests that SNPs explain between}

9.6-23.1% of familial risk.82

I

R

C

C

I

M

From a public health perspective, reducing overall risk and thereby preventing disease is extremely desirable. This can be achieved by using proven and acceptable prevention strategies.

P

Our comprehensive knowledge about risk and protective factors provides significant opportunity for primary prevention strategies to reduce the incidence of colorectal cancer in our population. As 45-60% of all colorectal cancers can be attributable to unhealthy lifestyle

factors,83-85 _{encouraging a healthy lifestyle, which encompasses a balanced diet and adequate}

physical activity, seems pertinent in this quest. As many of these risk factors are shared with other common chronic diseases, including diabetes and cardiovascular diseases, interventions to reduce their impact will provide the additional benefit to both individuals and the population as a whole.

In addition to encouraging a healthy, active lifestyle, there is evidence of effective chemoprevention of colorectal cancer using specific drugs. The best known chemoprevention

medications are non-steroidal anti-inflammatory drugs such as aspirin86 _{and hormone}

replacement therapy in postmenopausal women.87_{Long-term use (greater than five years) of}

at least 75 mg per day of aspirin has been demonstrated to reduce colorectal cancer risk.86

Although aspirin has the potential to lower colorectal cancer risk, usage may also result in serious adverse side effects including haemorrhagic strokes and gastrointestinal

complications such as peptic ulcers and bleeding.88_{Despite these risks, the balance of benefits}

to harms is considered favourable for those age 50 to 70 years at average risk of colorectal

cancer and aspirin should be considered as a preventative measure in this population.68, 89

Aspirin should be encouraged for individuals with Lynch syndrome and familial adenomatous

polyposis where surgery is inappropriate.68

S

/S

Secondary prevention focuses on the early detection of a disease to reduce disease-specific morbidity and mortality. For colorectal cancer, secondary prevention is predominantly achieved through screening. As the primary focus of this thesis is screening, it will be discussed in greater detail than other forms of prevention.

Screening is a public health intervention where asymptomatic individuals are tested for signs of a disease. The aim of screening is to detect a disease in its early stages, before symptoms develop and the disease spreads, as early detection generally increases the chances of successful treatment and survival. An effective and acceptable screening program should lead to mortality reductions without causing significant harm to the participants.

Colorectal cancer is especially suitable for screening and is one of a handful of cancers where a screening program has been proven to be effective at reducing incidence and mortality. It is an attractive and viable option satisfying most of the World Health Organisation’s criteria for

a cancer screening program.90_{Early colorectal cancer is often asymptomatic and may be}

present for several years before signs and symptoms become apparent. During this time, non-visible (occult) bleeding may occur and premalignant lesions can be removed before they become cancerous or colorectal cancers can be detected at earlier stages. This is beneficial for two reasons. Firstly, the identification and removal of premalignant lesions results in reduced

colorectal cancer incidence in the longer term,91_{the full effect of which may not been seen}

for several years.92 _{Secondly, detection of colorectal cancer in its early stages means the}

disease is more curable and therefore chances of survival are better.93-95_{Compared to those}

who present with symptoms, a higher proportion of early stage cancers are detected in those

who are screened.96-98

There are a range of acceptable, reliable and safe screening tools available for colorectal cancer screening. The most commonly used screening methods are stool-based tests and endoscopy methods. There are two types of stool-based occult blood tests, the guaiac faecal

In

tr

oduction

early age (below age 55) or two close relatives with colorectal cancer, irrespective of the age at diagnosis, colorectal cancer risk has been found to be three- to six-fold greater than the

average population.68 _{This risk increases to seven- to ten-fold when there are at least three}

first- or second-degree relatives with colorectal cancer, with at least one diagnosed under 55 years or at least three first-degree relatives with colorectal cancer diagnosed at 55 years or older.

In addition, there are several familial genes that substantially increase an individual’s risk of early-onset colorectal cancer when they are inherited in a mutated form. These genes are associated with Lynch syndrome (previously known as hereditary non-polyposis colorectal cancer), familial adenomatous polyposis (also known as FAP) and MUTYH-associated polyposis. However, together these inherited genetic mutations account for less than 6% of

all colorectal cancer cases69, 70 _{and do not fully explain why family history is a risk for colorectal}

cancer.71, 72 _{This suggests that other factors associated with heritability may play a role,}

including dietary, lifestyle and environmental factors,73 _{mutations in yet-to-be-discovered}

colorectal cancer susceptibility genes74 _{and common, low risk variants (or single-nucleotide}

polymorphisms [SNPs]).75-77 _{Although in isolation, SNPs are only weakly associated with}

colorectal cancer risk, individuals with multiple SNPs can have a substantially increased risk. In addition, they explain substantial variation in heritability risk due to their relatively high

prevalence in the population.77-81 _{Research to date suggests that SNPs explain between}

9.6-23.1% of familial risk.82

I

R

C

C

I

M

From a public health perspective, reducing overall risk and thereby preventing disease is extremely desirable. This can be achieved by using proven and acceptable prevention strategies.

P

Our comprehensive knowledge about risk and protective factors provides significant opportunity for primary prevention strategies to reduce the incidence of colorectal cancer in our population. As 45-60% of all colorectal cancers can be attributable to unhealthy lifestyle

factors,83-85 _{encouraging a healthy lifestyle, which encompasses a balanced diet and adequate}

physical activity, seems pertinent in this quest. As many of these risk factors are shared with other common chronic diseases, including diabetes and cardiovascular diseases, interventions to reduce their impact will provide the additional benefit to both individuals and the population as a whole.

In addition to encouraging a healthy, active lifestyle, there is evidence of effective chemoprevention of colorectal cancer using specific drugs. The best known chemoprevention

medications are non-steroidal anti-inflammatory drugs such as aspirin86 _{and hormone}

replacement therapy in postmenopausal women.87 _{Long-term use (greater than five years) of}

at least 75 mg per day of aspirin has been demonstrated to reduce colorectal cancer risk.86

Although aspirin has the potential to lower colorectal cancer risk, usage may also result in serious adverse side effects including haemorrhagic strokes and gastrointestinal

complications such as peptic ulcers and bleeding.88 _{Despite these risks, the balance of benefits}

to harms is considered favourable for those age 50 to 70 years at average risk of colorectal

cancer and aspirin should be considered as a preventative measure in this population.68, 89

Aspirin should be encouraged for individuals with Lynch syndrome and familial adenomatous

polyposis where surgery is inappropriate.68

S

/S

Secondary prevention focuses on the early detection of a disease to reduce disease-specific morbidity and mortality. For colorectal cancer, secondary prevention is predominantly achieved through screening. As the primary focus of this thesis is screening, it will be discussed in greater detail than other forms of prevention.

Screening is a public health intervention where asymptomatic individuals are tested for signs of a disease. The aim of screening is to detect a disease in its early stages, before symptoms develop and the disease spreads, as early detection generally increases the chances of successful treatment and survival. An effective and acceptable screening program should lead to mortality reductions without causing significant harm to the participants.

Colorectal cancer is especially suitable for screening and is one of a handful of cancers where a screening program has been proven to be effective at reducing incidence and mortality. It is an attractive and viable option satisfying most of the World Health Organisation’s criteria for

a cancer screening program.90 _{Early colorectal cancer is often asymptomatic and may be}

present for several years before signs and symptoms become apparent. During this time, non-visible (occult) bleeding may occur and premalignant lesions can be removed before they become cancerous or colorectal cancers can be detected at earlier stages. This is beneficial for two reasons. Firstly, the identification and removal of premalignant lesions results in reduced

colorectal cancer incidence in the longer term,91 _{the full effect of which may not been seen}

for several years.92 _{Secondly, detection of colorectal cancer in its early stages means the}

disease is more curable and therefore chances of survival are better.93-95 _{Compared to those}

who present with symptoms, a higher proportion of early stage cancers are detected in those

who are screened.96-98

There are a range of acceptable, reliable and safe screening tools available for colorectal cancer screening. The most commonly used screening methods are stool-based tests and endoscopy methods. There are two types of stool-based occult blood tests, the guaiac faecal

occult blood test (gFOBT) and the faecal immunochemical test (FIT). Although the tests are similar in nature in that they both detect small traces of blood in the stool and they can be completed at home, the tests differ significantly. A gFOBT requires sampling over multiple days, individuals are asked to modify their diet, including restriction of red meat intake, and some medications must be stopped during the stool collection phase. These restrictions are not necessary when completing a FIT. This is partly because gFOBT tests for the presence of any blood, whereas the FIT is specific for human blood. In addition, the FIT is a quantitative test which allows the choice of preferred cut-off for a positive test (µg Hb/g faeces). This is an important consideration as it impacts the balance between true and false positive test results and impacts demand for diagnostic follow-up. Several large randomised control trials (RCTs) have shown that annual or biennial screening with gFOBT will reduce colorectal cancer

mortality by 11-33%.93-95, 99-103 _{As the FIT is considered to be superior to gFOBT, it is expected}

that the mortality reduction will be larger when this test is used.104-106 _{While there are no RCTs}

to demonstrate this, the finding is supported by several observational studies.107-109

Two endoscopy methods are available, sigmoidoscopy and colonoscopy. Both methods inspect the colon using a flexible tube with a fibre optic camera. However, while colonoscopy inspects the entire colon, sigmoidoscopy only inspects the distal part. To date, only the effectiveness of sigmoidoscopy has been established with RCTs, with incidence reductions varying from 18-26% and mortality reductions from 22-31% (intention-to-treat analyses,

higher estimate were obtained for per-protocol analyses). 110-115 _{Because both tests are very}

similar, and colonoscopy visualizes the same segments as sigmoidoscopy and more, it is expected to be at least as effective as sigmoidoscopy. Randomised controlled trials of

colonoscopy screening are currently underway,116 _{but this expectation is already supported}

by evidence from observational studies.117-120

In addition to the aforementioned screening tests, there are a range of other test options and emerging technologies for detecting colorectal cancer. These include blood based biomarker

tests (Sept9121_{), imaging techniques (computed tomographic colonography}122 _and

double-contrast barium enema123_{), new endoscopy techniques (capsule endoscopy}124_{) and stool}

based tests incorporating DNA testing (multi-target stool DNA125_{). However, to date, these}

tests are rarely used in organised screening programs.

Every screening test has its advantages and disadvantages. Stool tests are the least invasive, and cheapest screening option available and can be performed at home. However, the major drawback of screening with stool-based tests (and sigmoidoscopy) is that they have to be followed by a colonoscopy for diagnosis and removal of lesions, plus the lack of sensitivity for stool tests means the tests need to be repeated frequently (annually or biennially). While colonoscopy is able to diagnose and remove lesions, as a screening tool it is not without shortcomings. Compared to stool-based tests, it is expensive and invasive, bowel preparation

is burdensome and the test is associated with rare but serious complications.126-130 _Although

the high sensitivity and specificity of the test means that the interval between repeated

screening events is substantially longer (generally 10 years),131 _{in many regions there is not}

enough capacity to screen all individuals in the target population with this test.

T

Tertiary prevention aims to reduce or prevent further complications and health impact and improve quality of life after a diagnosis with colorectal cancer. For individuals with a colorectal cancer diagnosis, tertiary prevention encompasses available treatment options and encouraging activities post treatment that can help reduce the risk of recurrent cancer. Significant advances have been made in colorectal cancer treatment over the last decades,

which has resulted in improved survivorship.37 _{The intensity, cost and effects of treatment}

depend on the stage and location of the cancer and may include surgery, radiotherapy, chemotherapy or a combination of these. Treatment of rectal cancer varies from that of colon cancer in terms of surgical technique, use of radiation therapy and the method of

administration of chemotherapy.132 _{Treatment is constantly evolving with new innovations}

and this, coupled with improvements in diagnostics, will lead to a continuous improvement in

the survival of colorectal cancer patients.37 _{However, such advancements come at}

considerable cost.

Several primary prevention factors have also been associated with improved outcomes and decreased risk of colorectal cancer-related death after a diagnosis and treatment of colorectal cancer. Maintaining a healthy bodyweight, being physically active, and eating a healthy diet can increase quality of life during chemotherapy and improve survival by reducing

cancer-specific and overall mortality by up to 40%.133, 134

G

S

C

C

S

The status of colorectal cancer screening varies widely around the world. Screening programs have predominantly been introduced in Western countries which generally have higher colorectal cancer incidence and more available resources. However, this is changing, with a

growing number of middle income countries introducing screening in recent years.135

Screening programs for colorectal cancer can be classified as either opportunistic or organised, or as a combination of the two. In opportunistic screening programs, like the US, Germany and Switzerland, screening occurs on an ad hoc basis, usually through fee-for service reimbursement of physicians. In an organised screening program, like Australia, the Netherlands and some regions of China, there is a systematic process which invites a target population to participate in screening and ensures follow-up of those with a positive screening

In

tr

oduction

occult blood test (gFOBT) and the faecal immunochemical test (FIT). Although the tests are similar in nature in that they both detect small traces of blood in the stool and they can be completed at home, the tests differ significantly. A gFOBT requires sampling over multiple days, individuals are asked to modify their diet, including restriction of red meat intake, and some medications must be stopped during the stool collection phase. These restrictions are not necessary when completing a FIT. This is partly because gFOBT tests for the presence of any blood, whereas the FIT is specific for human blood. In addition, the FIT is a quantitative test which allows the choice of preferred cut-off for a positive test (µg Hb/g faeces). This is an important consideration as it impacts the balance between true and false positive test results and impacts demand for diagnostic follow-up. Several large randomised control trials (RCTs) have shown that annual or biennial screening with gFOBT will reduce colorectal cancer

mortality by 11-33%.93-95, 99-103 _{As the FIT is considered to be superior to gFOBT, it is expected}

that the mortality reduction will be larger when this test is used.104-106 _{While there are no RCTs}

to demonstrate this, the finding is supported by several observational studies.107-109

Two endoscopy methods are available, sigmoidoscopy and colonoscopy. Both methods inspect the colon using a flexible tube with a fibre optic camera. However, while colonoscopy inspects the entire colon, sigmoidoscopy only inspects the distal part. To date, only the effectiveness of sigmoidoscopy has been established with RCTs, with incidence reductions varying from 18-26% and mortality reductions from 22-31% (intention-to-treat analyses,

higher estimate were obtained for per-protocol analyses). 110-115 _{Because both tests are very}

similar, and colonoscopy visualizes the same segments as sigmoidoscopy and more, it is expected to be at least as effective as sigmoidoscopy. Randomised controlled trials of

colonoscopy screening are currently underway,116 _{but this expectation is already supported}

by evidence from observational studies.117-120

In addition to the aforementioned screening tests, there are a range of other test options and emerging technologies for detecting colorectal cancer. These include blood based biomarker

tests (Sept9121_{), imaging techniques (computed tomographic colonography}122 _and

double-contrast barium enema123_{), new endoscopy techniques (capsule endoscopy}124_{) and stool}

based tests incorporating DNA testing (multi-target stool DNA125_{). However, to date, these}

tests are rarely used in organised screening programs.

Every screening test has its advantages and disadvantages. Stool tests are the least invasive, and cheapest screening option available and can be performed at home. However, the major drawback of screening with stool-based tests (and sigmoidoscopy) is that they have to be followed by a colonoscopy for diagnosis and removal of lesions, plus the lack of sensitivity for stool tests means the tests need to be repeated frequently (annually or biennially). While colonoscopy is able to diagnose and remove lesions, as a screening tool it is not without shortcomings. Compared to stool-based tests, it is expensive and invasive, bowel preparation

is burdensome and the test is associated with rare but serious complications.126-130 _Although

the high sensitivity and specificity of the test means that the interval between repeated

screening events is substantially longer (generally 10 years),131 _{in many regions there is not}

enough capacity to screen all individuals in the target population with this test.

T

Tertiary prevention aims to reduce or prevent further complications and health impact and improve quality of life after a diagnosis with colorectal cancer. For individuals with a colorectal cancer diagnosis, tertiary prevention encompasses available treatment options and encouraging activities post treatment that can help reduce the risk of recurrent cancer. Significant advances have been made in colorectal cancer treatment over the last decades,

which has resulted in improved survivorship.37 _{The intensity, cost and effects of treatment}

depend on the stage and location of the cancer and may include surgery, radiotherapy, chemotherapy or a combination of these. Treatment of rectal cancer varies from that of colon cancer in terms of surgical technique, use of radiation therapy and the method of

administration of chemotherapy.132 _{Treatment is constantly evolving with new innovations}

and this, coupled with improvements in diagnostics, will lead to a continuous improvement in

the survival of colorectal cancer patients.37 _{However, such advancements come at}

considerable cost.

Several primary prevention factors have also been associated with improved outcomes and decreased risk of colorectal cancer-related death after a diagnosis and treatment of colorectal cancer. Maintaining a healthy bodyweight, being physically active, and eating a healthy diet can increase quality of life during chemotherapy and improve survival by reducing

cancer-specific and overall mortality by up to 40%.133, 134

G

S

C

C

S

The status of colorectal cancer screening varies widely around the world. Screening programs have predominantly been introduced in Western countries which generally have higher colorectal cancer incidence and more available resources. However, this is changing, with a

growing number of middle income countries introducing screening in recent years.135

Screening programs for colorectal cancer can be classified as either opportunistic or organised, or as a combination of the two. In opportunistic screening programs, like the US, Germany and Switzerland, screening occurs on an ad hoc basis, usually through fee-for service reimbursement of physicians. In an organised screening program, like Australia, the Netherlands and some regions of China, there is a systematic process which invites a target population to participate in screening and ensures follow-up of those with a positive screening

test.135 _{Although opportunistic and organised screening can yield similar uptake rates,}

organised programs have a greater potential to reduce cancer incidence and mortality, due to higher levels of population coverage and a centralised commitment to quality and

monitoring.136, 137 _{Moreover, the focus on quality assurance through each step in the}

organised screening process provides greater protection against the possible harms of screening, inappropriate use of resources and poor follow-up of those with a positive screen.

In this way, organised screening programs are also more likely to be cost-effective.137

The different screening programs throughout the world reflect the uncertainty about which strategy is best and differences in decision-making processes. The decision to implement a screening program is a multi-faceted and complex process. The effectiveness of a screening program relies heavily on participation, which is largely determined by population

preference.132 _{Among other factors, participation in screening is affected by the expected and}

perceived burden of the test, the risk of complications and other harms and cultural beliefs of

the screened individual.135 _{Other than population preference, the choice of test needs to take}

into account several other aspects including: test sensitivity and specificity, resource capacity, costs and the harm-benefit ratio. In addition, screening starting age, stopping age and interval need to be considered.

M

U

While indispensable, traditional methods of gathering empirical evidence of screening effectiveness, such as randomised control trials, case control and cohort studies, are not always feasible and are not without limitations. Although such methods are considered to provide accurate and reliable evidence, they do so under specific conditions that often differ from future daily application and may not be generalisable to other settings. In addition, such methods can only evaluate a limited number of interventions at a time, are expensive and time consuming – requiring several years of follow-up before the effectiveness (colorectal cancer incidence or mortality reduction) of the intervention can be evaluated. Finally, traditional methods usually have a limited follow-up time and as such they are unable to determine lifetime health effects and costs, which is necessary to determine the (cost-) effectiveness of screening. Moreover, as screening strategies can vary in a multitude of ways, with different screening tests, age ranges, screening intervals and referral threshold for follow-up testing, the options are so numerous that it is impossible to compare them all in the traditional way.

Decision models have been developed to allow for the evaluation of many different screening strategies – any number of screening strategies can be simulated and results generated in a

short time frame. Furthermore, models can be adjusted to reflect the setting of interest, taking into account colorectal cancer risk, life expectancy, resource availability and population preferences. These assumptions can also be adjusted to allow comparison of how outcomes of interest might change. Decision models therefore provide a useful tool to extrapolate evidence from traditional methods and address the question of which screening strategy is

optimal given local conditions.138 _{As such, models have great potential to assist in the}

development, guidance and decision making process of public health initiatives, including

colorectal cancer screening programs.139, 140

In this thesis we have used two models to assist in answering policy questions for optimising colorectal cancer screening. The main model used is the MISCAN-Colon model which is

described in more detail in the following section and in the Model Appendix. To establish the

costs and effects of Lynch syndrome screening in the Australia, we developed a second,

simplified, decision analysis model (Chapter 1).

MISCAN-C

Microsimulation Screening Analysis-Colon (MISCAN-Colon) is a well-established microsimulation model for CRC developed at the Department of Public Health, Erasmus

University Medical Center.141 _{In brief, the model first simulates the life histories of a}

hypothetical population of individuals from birth to death without screening for colorectal cancer. As the simulated individuals age, adenomas may arise, some may progress in size and some may develop into cancer. During each stage symptoms may present and a colorectal cancer diagnosed. The model then simulates the same population with screening. The introduction of screening potentially alters the simulated life histories through detection and removal of adenomas or through detection of colorectal cancer at earlier stages. By comparing the life histories of a simulated population being screened to the simulated population not screened, MISCAN-Colon quantifies the effectiveness and the costs of screening.

MISCAN-Colon is made of three components which consider demography assumptions, natural history assumptions, and screening assumptions (Figure 3). Where possible these assumptions are derived from literature, however, some natural history and screening assumptions are not readily available or are uncertain. When this is the case, model calibration is required to estimate these parameters or to reduce uncertainty. A more detailed

In

tr

oduction

test.135 _{Although opportunistic and organised screening can yield similar uptake rates,}

organised programs have a greater potential to reduce cancer incidence and mortality, due to higher levels of population coverage and a centralised commitment to quality and

monitoring.136, 137 _{Moreover, the focus on quality assurance through each step in the}

organised screening process provides greater protection against the possible harms of screening, inappropriate use of resources and poor follow-up of those with a positive screen.

In this way, organised screening programs are also more likely to be cost-effective.137

The different screening programs throughout the world reflect the uncertainty about which strategy is best and differences in decision-making processes. The decision to implement a screening program is a multi-faceted and complex process. The effectiveness of a screening program relies heavily on participation, which is largely determined by population

preference.132 _{Among other factors, participation in screening is affected by the expected and}

perceived burden of the test, the risk of complications and other harms and cultural beliefs of

the screened individual.135 _{Other than population preference, the choice of test needs to take}

into account several other aspects including: test sensitivity and specificity, resource capacity, costs and the harm-benefit ratio. In addition, screening starting age, stopping age and interval need to be considered.

M

U

While indispensable, traditional methods of gathering empirical evidence of screening effectiveness, such as randomised control trials, case control and cohort studies, are not always feasible and are not without limitations. Although such methods are considered to provide accurate and reliable evidence, they do so under specific conditions that often differ from future daily application and may not be generalisable to other settings. In addition, such methods can only evaluate a limited number of interventions at a time, are expensive and time consuming – requiring several years of follow-up before the effectiveness (colorectal cancer incidence or mortality reduction) of the intervention can be evaluated. Finally, traditional methods usually have a limited follow-up time and as such they are unable to determine lifetime health effects and costs, which is necessary to determine the (cost-) effectiveness of screening. Moreover, as screening strategies can vary in a multitude of ways, with different screening tests, age ranges, screening intervals and referral threshold for follow-up testing, the options are so numerous that it is impossible to compare them all in the traditional way.

Decision models have been developed to allow for the evaluation of many different screening strategies – any number of screening strategies can be simulated and results generated in a

short time frame. Furthermore, models can be adjusted to reflect the setting of interest, taking into account colorectal cancer risk, life expectancy, resource availability and population preferences. These assumptions can also be adjusted to allow comparison of how outcomes of interest might change. Decision models therefore provide a useful tool to extrapolate evidence from traditional methods and address the question of which screening strategy is

optimal given local conditions.138 _{As such, models have great potential to assist in the}

development, guidance and decision making process of public health initiatives, including

colorectal cancer screening programs.139, 140

In this thesis we have used two models to assist in answering policy questions for optimising colorectal cancer screening. The main model used is the MISCAN-Colon model which is

described in more detail in the following section and in the Model Appendix. To establish the

costs and effects of Lynch syndrome screening in the Australia, we developed a second,

simplified, decision analysis model (Chapter 1).

MISCAN-C

Microsimulation Screening Analysis-Colon (MISCAN-Colon) is a well-established microsimulation model for CRC developed at the Department of Public Health, Erasmus

University Medical Center.141 _{In brief, the model first simulates the life histories of a}

hypothetical population of individuals from birth to death without screening for colorectal cancer. As the simulated individuals age, adenomas may arise, some may progress in size and some may develop into cancer. During each stage symptoms may present and a colorectal cancer diagnosed. The model then simulates the same population with screening. The introduction of screening potentially alters the simulated life histories through detection and removal of adenomas or through detection of colorectal cancer at earlier stages. By comparing the life histories of a simulated population being screened to the simulated population not screened, MISCAN-Colon quantifies the effectiveness and the costs of screening.

MISCAN-Colon is made of three components which consider demography assumptions, natural history assumptions, and screening assumptions (Figure 3). Where possible these assumptions are derived from literature, however, some natural history and screening assumptions are not readily available or are uncertain. When this is the case, model calibration is required to estimate these parameters or to reduce uncertainty. A more detailed